Closed YangNuoCheng closed 2 years ago
@YangNuoCheng You can use QKeras to train quantized models that can be mapped to fixed-precision types and deployed on FPGAs. Once such project with (almost) full compatibility with QKeras is hls4ml. Disclaimer: I'm one of the developers of that project.
This work https://rdcu.be/cmVO7 (published in Nature Machine Intelligence) used QKeras + hls4ml on FPGA.
I find the 'fpga-accelerator' tag, but there's no FPGA in repository.